The proliferation of transistors has increased the capacity and performance of computing systems by over a factor of a million in the past thirty years. One corollary of Moore's Law is the dramatically increasing amount of data in existence, driven by improvements in sensor, communication, and storage technology. One example of such data is observational data, such as satellite imagery data. Unfortunately, factors such as the sheer size of observational datasets, and the evolution of file formats for storing such datasets over the years, pose substantial logistical and computational challenges to entities wishing to use such datasets. Accordingly, there is an ongoing need for systems and techniques capable of handling large sets of observational data.